NEW PERK: WAR Roster Builder 2.0 (Patreon)

It's a long time coming, but the update to the WAR Roster Generator (now Builder) is here. 

I've sent invites to the Patreon email list, but if you didn't receive one please request access here: (Link Temporarily Taken Down - See Post Here)

The main new feature here is the 2020-21 Depth Chart sheet, which allows you to select a team and automatically inputs their current depth chart (based on the CapFriendly projection). This gives you the ability to see how current rosters stacked up based on their projected WAR, including how they rank compared to the rest of the league in terms of forwards, defencemen, goalies, and overall. 

Another addition is the inclusion of both offensive and defensive line- and position-level projections, so you can see how certain units fare at both ends of the ice. For example, here's the Edmonton Oilers' forward group. 

You can see that while their top six is excellent offensively, it's also real rough defensively. This wouldn't have been clear from the original roster generator.  

To change the depth charts and line combinations, you will have to Make a Copy of the spreadsheet and go into "Line Ref." From there you will be able to edit the lines as you see fit. Players who did not play 200 minutes in 2019-20 are represented as replacement-level, including rookies. Special teams are carried over from the previous year, as there is not a reliable way to project it outwards.

Also added is the ability to change goaltender workload in order to solve the issue that all starting goalies were previously projected as playing 55 games. This is connected to the rankings as well, so if there's a shift it will be reflected in the team's point projection.

Speaking of projections, I've ironed out quite a few issues with them. The biggest one is that there were a lot of projections that were massively influenced by shooting luck, because WAR is based on driving goals for and not expected goals for. This led to some very bizarre projections, such as Jared McCann being rated higher than Evgeni Malkin and Sidney Crosby.  To ameliorate this, I analyzed differences between WAR and xWAR to pick out seasons that were clearly inflated by luck (as opposed to players who consistently outperform expected goals) and revert them to xWAR. While there are still plenty of players whose superb performance in low-minute roles gives them big WAR/60 - it will be up to you guys to recognize that Nic Dowd would probably not be a superstar if played on line one.

Something you might have noticed is that individual player WAR has been left out of the 2020-21 projected slides. This was a conscious decision made in the interest both of protecting the data from Evolving-Hockey.com and removing the distraction of "x player is better than y player" which inevitably dominate the conversation around these types of visualization. The 1.0 version of the roster builder is still available if you really want that element.

Enjoy, and let me know if you notice any glitches, bugs, or other issues.